English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

To model the indeterminacy of human behaviors, stochastic trajectoryprediction requires a sophisticated multi-modal distribution of futuretrajectories. Emerging diffusion models have revealed their tremendousrepresentation capacities in numerous generation tasks, showing potential forstochastic trajectory prediction. However, expensive time consumption preventsdiffusion models from real-time prediction, since a large number of denoisingsteps are required to assure sufficient representation ability. To resolve thedilemma, we present LEapfrog Diffusion model (LED), a novel diffusion-basedtrajectory prediction model, which provides real-time, precise, and diversepredictions. The core of the proposed LED is to leverage a trainable leapfroginitializer to directly learn an expressive multi-modal distribution of futuretrajectories, which skips a large number of denoising steps, significantlyaccelerating inference speed. Moreover, the leapfrog initializer is trained toappropriately allocate correlated samples to provide a diversity of predictedfuture trajectories, significantly improving prediction performances. Extensiveexperiments on four real-world datasets, including NBA/NFL/SDD/ETH-UCY, showthat LED consistently improves performance and achieves 23.7%/21.9% ADE/FDEimprovement on NFL. The proposed LED also speeds up the inference19.3/30.8/24.3/25.1 times compared to the standard diffusion model onNBA/NFL/SDD/ETH-UCY, satisfying real-time inference needs. Code is available athttps://github.com/MediaBrain-SJTU/LED.

Leapfrog Diffusion Model for Stochastic Trajectory Prediction